phase 90 with CD4009?

[carlozsulca]

hello, recently I made a thread on a tap lfo with the attiny84 for a VB-2 and then when I finished building it, I came up with the idea of making a phaser with tap tempo, I would like to make a small stone but to start with the phaser world I want to start with phase 90 of mxr, I was looking at the schematics and they have about 2n5952 and they must be matched and that transistors are no longer found and the pairs that they sell on ebay are very expensive, investigating a little more I found the schematic of the EHX bad stone and the old version looks a lot like the phase 90 design but in the modern one they changed those transistors for a CD4009, my question is: can these transistors be replaced by a CD4009 in phase 90 as they did in the bad stone?

[Rob Strand]

It would be a major job to replace JFETs with MOSFET (CMOS gates).   A redesign based on
the Bad Stone.    You might be able to change the LFO from triangular to exponential
like the phase 90 to make it more like the phase 90.

A CD4049 seems the way to go.

The way the CD4009 is connected in the Bad Stone should work with a CD4049.

https://www.diystompboxes.com/smfforum/index.php?topic=127015.0

There are differences between the CD4049 and CD4009 devices.  The Bad Stone connections
accommodate those differences.

Technically it might be better to leave pin 16 of the CD4049 disconnected.
Pin 16 is NC on the CD4049 but on the CD4009 it is the Vdd pin (power to inputs and input diodes).


FYI:

Putting a CD4009 into a CD4049 phaser, like the ETI  447 Phaser,  is likely to cause problems because
the ETI is taking advantage of the CD4049's lack of input diodes to Vdd.    The Bad Stone connect is more clever
because they managed to get it to work without that luxury.

There are some weird points about both the ETI 447 and the Bad Stone.   Pin 13 is connected but it is an NC pin
for both the CD4049 and CD4009.   Perhaps one of the brands and chips has some fine print in the datasheet.

[Rob Strand]

2n5952

If you are worried about getting that particular JFET, don't be.  There's plenty of other JFETs which will work in that circuit.  Many I'm sure you can buy pre-matched sets for on-line.

[Radical CJ]

The Pearl 3 apparently uses a CD4069, although the schematic kicking around the net shows it as "proprietary chip".

The issue is that the LFO in the Pearl 3 looks very complicated compared to a Phase90 and the large number of surrounding components make it discouraging to breadboard as an experiment (it also has has a very nifty bypass system useing a CD4007, which makes the while schematic look pretty intimidating).

In the back of my mind I have been wondering if their is an easier implementation of a CD4069 phaser, that is more diy friendly. However, my current project is actually trying to get a Phase45-alike phaser working with another unconventional replacement for the JFETs. Once it works well enough I'll post a schematic, but at the moment it sounds more like a Univibe.

[Rob Strand]

The Pearl 3 apparently uses a CD4069, although the schematic kicking around the net shows it as "proprietary chip".

The issue is that the LFO in the Pearl 3 looks very complicated compared to a Phase90 and the large number of surrounding components make it discouraging to breadboard as an experiment (it also has has a very nifty bypass system useing a CD4007, which makes the while schematic look pretty intimidating).

In the back of my mind I have been wondering if their is an easier implementation of a CD4069 phaser, that is more diy friendly. However, my current project is actually trying to get a Phase45-alike phaser working with another unconventional replacement for the JFETs. Once it works well enough I'll post a schematic, but at the moment it sounds more like a Univibe.

I can confirm the Pearl 03 uses a MC14069UB.   Some pics of boards were put up clearly showing the non-custom chip.

The Pearl is a 4-stage phaser.   Four of the six  4069 gates are used as MOSFET controlled resistors.   One of the "free" 4069's was used to provide a bias to match the other gates.   It seem one gate is not used, I can't see the connections but it should connect somewhere.

The 4007 gate is used for the footswitch and MOSFET audio switching.

It's a clever circuit.

The LFO is a filtered triangle-wave.  So kind of sine-ish.

[Radical CJ]

There are some weird points about both the ETI 447 and the Bad Stone.   Pin 13 is connected but it is an NC pin
for both the CD4049 and CD4009.   Perhaps one of the brands and chips has some fine print in the datasheet.

I'd never seen the ETI 447 before, this looks like a good starting point for an easier implementation of the PH 3's CD4069 approach. I've starting putting it together on a breadboard, I think my plan will be to:

Keep the 6 modulated stages and add 2 unmodulated stages
Add feedback/regeneration loop
Replace JFET input and BJT output with opamps
Test whether the LFO can accommodate / \ ramps

[Rob Strand]

I'd never seen the ETI 447 before, this looks like a good starting point for an easier implementation of the PH 3's CD4069 approach. I've starting putting it together on a breadboard, I think my plan will be to:

The ETI 447 takes advantage of the 4049 and probably ends up with the most minimalistic circuit, although maybe not the best.

There's a bit of a trade-off with noise and distortion with those CMOS gate designs.   You will notice the level is reduced at the input of the all-pass filters then boosted at the end.  I don't know how much improvement the PH-03 gives in this area.    The ETI probably has too much attenuation and the use of LM741's doesn't help noise.

FYI another simple one was the Ampeg Phazzer.  Someone put up the schematic within the last year or so.   It uses a CD4069 and gets around some of the issues with the CMOS input and output diodes by simply letting the CMOS gate + supply (+Vdd) float.    It should work.   The Phazzer  follows a simple circuit structure like the ETI, in fact it is very much along the lines of the MXR phase 90  - except for the LFO which is triangular like the ETI.    IIRC, the Phazzer doesn't use any input attenuation before the all-pass filters, which might be going too far as well.

[Radical CJ]

...   You will notice the level is reduced at the input of the all-pass filters then boosted at the end.  I don't know how much improvement the PH-03 gives in this area.    The ETI probably has too much attenuation and the use of LM741's doesn't help noise.

FYI another simple one was the Ampeg Phazzer.  Someone put up the schematic within the last year or so.   It uses a CD4069 and gets around some of the issues with the CMOS input and output diodes by simply letting the CMOS gate + supply (+Vdd) float.

Yes, I did notice the filtered audio path goes through a 10k/1k voltage divider, and then volume parity is restored at the mixing end with 5k6 and 56k resistors for the wet and dry signals respectively.

Easy enough to play with values once the circuit is working to see how much volume drop is actually needed. And of course I'm using TL074s, which might help with other noise issues.

I'll definitely take a look at the Phazzer, and experiment to see if the trick of floating vdd+ is an improvement.

Probably won't end up with the world's best phaser, but I'm really just curious to see what I can get to work really.

I've got 2 phaser questions someone might be able to help with.

1) is there a formula (or online calculator) for calculating where the notches fall? The ETI artical provides example resistor and cap values and their respective notches, but doesn't show the maths. And I found a link to a calculator on an old thread here, but it was long dead (404). It's something I would have thought the synth people would use, but I couldn't find anything on their forums either.

2) are there any particular LFO circuits that are considered great for phaser? The LFOs in the classic 2 and 4 stage phasers all look very basic. But if triangle sounds best then there's no point trying to create a pseudo sinewave I guess.

Much obliged,

[Rob Strand]

1) is there a formula (or online calculator) for calculating where the notches fall? The ETI artical provides example resistor and cap values and their respective notches, but doesn't show the maths. And I found a link to a calculator on an old thread here, but it was long dead (404). It's something I would have thought the synth people would use, but I couldn't find anything on their forums either.

The notches occur at, f = k / (2 * pi * RC)
For a 4-stage phaser you get two notches, k = 0.414 and k = 2.414.
For a 6-stage phaser you get three notches, k = 0.267, k = 1, k = 3.732
You can see the 8-stage on wikipedia, roughly k = 0.2, 0.66, 1.5, 5.0
Notice the notches are always symmetrical about k=1, in that 1/0.414 = 2.414 and 1/0.267 = 3.732.

C = all pass cap value and R = JFET or MOSFET resistance which is variable.

2) are there any particular LFO circuits that are considered great for phaser? The LFOs in the classic 2 and 4 stage phasers all look very basic. But if triangle sounds best then there's no point trying to create a pseudo sinewave I guess.

They all have their own sound.   The most even sounding sweep is more like a rectified sine-wave.  A bit like the Ross Phaser with the OTA based LFO.   I think the EHX bad-stone and the Mutron use a diode to get a roughly similar shape.   if you listen to an MXR sometimes they can sound even.  It's a weird thing when you start putting LFO's side by side the unevenness stands out more.   Many moons ago I did a lot of experiments with Phasers in DSP.   Then you have the univibes which are very much uneven sweeps but the pulsation effect stills sounds good!

[Radical CJ]

The notches occur at, f = k / (2 * pi * RC)
For a 4-stage phaser you get two notches, k = 0.414 and k = 2.414.
For a 6-stage phaser you get three notches, k = 0.267, k = 1, k = 3.732
You can see the 8-stage on wikipedia, roughly k = 0.2, 0.66, 1.5, 5.0
Notice the notches are always symmetrical about k=1, in that 1/0.414 = 2.414 and 1/0.267 = 3.732.

C = all pass cap value and R = JFET or MOSFET resistance which is variable.

Thanks for the math!